This action might not be possible to undo. Are you sure you want to continue?
Dynamics of an activated sludge process with nitriﬁcation and denitriﬁcation: Start-up simulation and optimization using evolutionary algorithm
Saziye Balku, Ridvan Berber ∗
Department of Chemical Engineering, Faculty of Engineering, Ankara University Tandogan, 06100 Ankara, Turkey Received 3 June 2004; received in revised form 3 October 2005; accepted 6 October 2005 Available online 28 November 2005
Abstract An alternative but beneﬁcial way of achieving nitriﬁcation and denitriﬁcation in activated sludge systems is to apply consecutive aeration and non-aeration periods in a single reactor. In this paper, dynamic modeling and start-up simulation in order to establish the continuous operation mode in such systems are ﬁrst presented. Newly developed ASM3 is applied to the modeling of microbiological processes in the aeration tank and a 10 layer settling model is adapted to the settling tank. Then, the constrained optimization problem is tackled by an evolutionary algorithm, and the optimum aeration schedule for minimum energy consumption is identiﬁed. The constraint handling is achieved by three comparative methods using rejection of infeasibles, penalizing infeasibles and tournament selection without using any penalty parameters. The results indicate that the suggested optimum operation strategy may be economically beneﬁcial in all methods but the third one, which is proposed by Deb [Deb, K. (2000). An efﬁcient constraint handling method for generic algorithms. Computer Methods in Applied Mechanics and Engineering, 186, 311–338] gives better solutions. © 2005 Elsevier Ltd. All rights reserved.
Keywords: Wastewater treatment; Dynamic optimization; Evolutionary algorithm
1. Introduction Activated sludge process, one of the ways of biological treatment, is commonly used in domestic wastewater treatment, or in the secondary treatment of the industrial wastewater. In the classical biological treatment systems anoxic denitriﬁcation is maintained in a separate zone, or carrousel type ditches are used for this purpose. In recent years alternating activated sludge processes have been developed in a way through which the aerobic and anoxic zones are periodically exchanged (Zhao, Isaacs, Soeberg, & K¨ ummel, 1995). In the present study a single aeration tank has been used for both aerobic and anoxic operation by an on/off strategy applied to the aeration devices determined by dynamic optimization. Such a system using aeration and non-aeration modes periodically in the aeration tank will ensure carbon removal, nitriﬁcation
Corresponding author. Tel.: +90 312 212 2130; fax: +90 312 212 7464. E-mail addresses: firstname.lastname@example.org (S. Balku), email@example.com (R. Berber). 0098-1354/$ – see front matter © 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.compchemeng.2005.10.014
and denitriﬁcation. The system, which can also be named as completely stirred alternating aerobic–anoxic (AAA) process, consists of basically an aeration tank and a settler as shown in Fig. 1. In the aeration tank, the microorganisms are used to convert the colloidal and dissolved carbonaceous organic matter into various gases and into cell tissue. Because cell tissue has a speciﬁc gravity slightly greater than that of water, the resulting cells can be removed from the treated liquid by gravity settling in the settler. Most of the cells are recycled and mixed with incoming wastewater to maintain convenient sludge age characteristics and such high degradation rates. In activated sludge systems, the main operational cost is due to the electrical energy needed to operate the aerating devices in order to obtain necessary aeration and mixing. Determining the optimum durations of consecutive aeration and non-aeration periods in order to minimize the energy consumption is, therefore, a non-trivial dynamic optimization problem. The dynamical character comes from the fact that the complicated dynamic model, which forms the constraints, needs to be satisﬁed at every step. Furthermore, the efﬂuent from the plant has to comply with
In the solution of the optimization . 1996). nitrifying biomass XSS total suspended solids strict discharge regulations. Gujer. In this perspective. this paper describes the dynamic model. which corrects some defects in ASM1 that have been noticed over 10 years of experience. (2001). nitriﬁcation and denitriﬁcation. are: (i) the storage of poly-hydroxy-alkanoates observed under aerobic and anoxic conditions is included in ASM3. in fact only one. reported in the current literature (Chachuat. Therefore. Some important features that ASM3 covers. Similar activated sludge systems (i. 2001). and attempts to deﬁne an effective and practical method to reach a solution by using more realistic means. alternating aerobic–anoxic process) have been studied from the perspective of process evaluation (Hao & Huang. Their dynamic model was constructed on ASM1 with some revisions including pH and alkalinity. Schematic diagram of the activated sludge process. sludge production. the problem of constrained optimization is tackled to ﬁnd the optimum aeration schedule for minimum energy consumption. 1. Thus. which was based upon ASM1 (Henze.e. & Matsuo. Since the growth of ﬁlamentous bacteria may be possible and also the growth rate and the activity of the nitrifying organisms decrease at low DO concentrations. compared to ASM1. Balku. 1996). it has become a new standard for future modeling. & Loosdrecht. Although ASM3 is based on the same dominating phenomena in terms of the oxygen consumption. Berber / Computers and Chemical Engineering 30 (2006) 490–499 491 Nomenclature abs at ave error in COD DO eff fns h J kL a Q R req rh rp absolute aeration average deviation from discharge criteria inﬂuent chemical oxygen demand dissolved oxygen efﬂuent non-settleable fraction of the inﬂuent suspended solids height gravity ﬂux oxygen transfer coefﬁcient ﬂow rate appearance/disappearance rates of components required settling parameter associated with the hindered settling component settling parameter associated with the low concentration and slowly settling component of the suspension returning sludge saturation settler total suspended solids total operation time total nitrogen tank volume maximum theoretical settling velocity maximum practical settling velocity settling velocity waste sludge Fig. and dynamic modeling (Huang & Hao. no quantitative assessment was made because they looked at the case primarily from the perspective of parameter estimation and process control. However. & Latiﬁ. there seems to be very limited work. 1999). Initially. Marais. (ii) the decay rates of nitriﬁers and heterotrophs under aerobic and anoxic conditions are separated and (iii) all the conversion processes of the two groups of organisms are clearly separated. R.S. Roche. the work tackles not only an academic but also an industrially important optimization problem. rs sat set SS T TN V v0 v0 VS w ASM3 components SO dissolved oxygen concentration SI inert soluble organic material SS readily biodegradable organic substrate SNH ammonium plus ammonia nitrogen S N2 dinitrogen SNO nitrate plus nitrite nitrogen SHCO alkalinity XI inert particulate organics slowly biodegradable substrates XS XH heterotrophic biomass XSTO organics stored by heterotrophs XA autotrophic. Grady. 1987). and this makes the hard constraints. the dissolved oxygen (DO) constraints are added to the optimization problem in order to avoid obtaining misleading results. the question remains open except the previous attempt by Chachuat et al. ASM3 is considered as the best choice in the modeling of the biological processes involved in the activated sludge tank for the time being especially in dealing with the aerobic and anoxic processes in the same tank. It was pointed out that the energy saving could be possible by the reducing aeration time in the steady state process evaluation. Henze. Furthermore. Mino. Finally. and provides the start-up simulation in order to establish the continuous operation mode. The fact that dynamic optimization problems are inherently difﬁcult to solve in the ﬁrst place makes the problem even more challenging. The present work uses the newly developed ASM3 (Gujer.
total ﬂux is equal to the sum of the bulk and gravity ﬂuxes.492 S. min(v0 . the mass balance related to dissolved oxygen includes an additional term on sat − S at )) representing the oxygen the right hand size (+kL a(SO O transfer. In the aeration period. 2. The dynamical work previously reported in the literature was based on ASM1. it is proposed to use the evolutionary algorithm. and their gravity ﬂuxes are set to zero. Mass balances in the aeration tank result in: at in + Q Xrs − (Q + Q )Xat dXi Qin Xi rs i rs in i + Ri = dt Vat concentrations in each layer. Berber / Computers and Chemical Engineering 30 (2006) 490–499 problem. In the settling tank. 3 was used for the biological processes involved in the aeration tank. Xrs and Xin are 13-dimensional vectors consisting of ASM3 components. Settling tank The settler is modeled as a cylindrical tank with 10 horizontal layers and settling velocity of Tak´ acs is used (Tak´ acs. The methods applied and results obtained by these methods are also compared in this study. For the particulate components of ASM3.90(XH + XA ) + 0. (1991). each layer is assumed to be perfectly stirred. and therefore reports an improvement in this context. Aeration tank Activated Sludge Model No. v0 (e−rh (Xt VS (Xt set −f Xset ) ns in −e−rp (Xt set −f Xset ) ns in ))) (2) where v0 . The results show that carbon removal. which has become popular in recent years. One of the main advantages of evolutionary computational techniques is that they do not have much mathematical requirements about the optimization problem. Xset ) dt where Xset is a 60-dimensional vector. i. Balku. whose concentrations are denoted by X.81% in comparison to an optimized constant proﬁle. The presence of the constraints makes the solution of the problem rather difﬁcult and three methods of dealing with constraints are proposed here. 2000). stoichiometric matrix and kinetic parameters (at 20 ◦ C) were taken from ASM3 as they were. The components of the system. Hence. despite discrimination for the aerobic and anoxic cycles in the aeration tank.60XSTO Thus. 2.e. include the equations for the settling tank since there is a process of recycle from the settling tank to the aeration tank. Differential equations related to the mass balances deﬁned by the ﬂux theory for each 10 layer and each particulate component then becomes: dXset = fset (Xat . concentrations of which are denoted by S. are assumed to be associated with the activated sludge concentrated in the settling tank. 3. but the problem has been deﬁned here by making use of a more realistic model. are divided into two categories.75(XI (3) + XS Xt ) + 0.1. 2. whereas particulate components. 1991). state variables. According to the ﬂux theory. nitriﬁcation and denitriﬁcation can be well achieved by the optimal aeration proﬁle determined by using evolutionary algorithm and 18. Parameters involved in settling . Patry. Both sets. No reaction takes place and no radial concentration changes. the rejection of infeasible individuals. 2. penalizing infeasibles and Deb’s method (Deb.98% energy saving is possible in comparison to a constant arbitrary aeration proﬁle and 5. the settling velocities should be calculated for each layer of the settler. all they need is an evaluation of the objective function. rp and fns are model parameters deﬁned in set the suspended solids concentrations Tak´ acs et al. The bulk ﬂuxes are the functions of the linear ﬂow rate and the solid where X is a 73-dimensional vector. the concentrations of soluble components of ASM3 are equal in the inlet and outlet of the settling tank. soluble components. The inlet wastewater composition. The soluble components of ASM3 are assumed to follow the water ﬂow. In order to calculate the gravity ﬂuxes. General model The general model then consists of two sets of differential equations as follows: dX = f (1) (X) dt dX = f (2) (X) dt non-aeration periods aeration periods (6) (7) (5) (1) where Xat .2. the gravity ﬂux for the particulate components at each layer j is: set (j ) J (j ) = VS (j )Xt (4) The minimum ﬂuxes and the threshold value for solid concentration are considered in the mass balances around each layer of the settler. Dynamic simulation Simulation of the start-up period of the plant was accomplished by starting with an aeration tank and a settler ﬁlled in by the incoming wastewater.3. R. Xin set is the concentration of the suspended entering the settler and Xt solids calculated for each layer of the settler based upon ASM3 as follows: set set set set set set = 0. the settling velocity is derived from Tak´ acs’ expression as follows: set ) = max(0. Treatment plant model Treatment plant model consists of an aeration tank model in which the microbiological processes take place and a settling tank model where the settling velocity of activated sludge is considered. & Nolasco. 13 aeration tank variables and 60 settling tank variables. v0 . are assumed to be transported by water. rh .
1 0.0 75.3 3. The threshold concentration Xt was equal to 3000 g m−3 .4 3.0 0.9 h nonaerated plus 1. In the recycling stream the concentrations of SO and SN2 were assumed to be equal to zero.3 – 6.0 25.7 1414.5 h−1 Simulation program was implemented in MATLAB by using ode15s integration routine for stiff differential equations with adjustable step size.3 – 6. R.7 3312.4 0. To maintain a balance between computing accuracy and time.2 8. A constant arbitrary aeration sequence of 0. Alkalinity shows a slight and continuous decrease in the start-up period. The changes in the concentrations of ASM3 components in the aeration tank as a function of time can be followed in the ﬁgures. 3. autotrophics (XA ).3 3.0 0.S.5 m kL a: 4. During the start-up period Table 1 The simulated concentrations of ASM3 components ASM3 components (g m−3 ) SO SI SS SNH SN2 SNO SHCO (moles m−3 ) XI XS XH XSTO XA XSS Initial values of aeration tank 0. The results of the simulation for 20 days of start-up operation for the system are shown in Table 1. slowly biodegradables (XS ).8 h aerated period was employed for 20 days. Twenty days simulation proﬁles for organics stored (XSTO ). 2) and the autotrophic bacteria concentration increases from 0. Fig.3 0.1 Sludge from settler bottom and recycle – 30.0 5.3 3.01 h for start-up simulation.1 125.0 0. the maximum step size was 0. ammonium nitrogen and nitrate nitrogen and also the alkalinity are shown in Fig.1 3. and also to achieve the initial concentrations for the system before the continuous operation mode for which the optimum aeration sequence was to be obtained. 1991) as follows: Qin : 1000 m3 day−1 Qrs : 800 m3 day−1 Vat : 450 m3 Vset : 400 m3 hset : 3. inerts soluble (SI ) and readily biodegradables (SS ). 2. Twenty days simulation proﬁles for suspended solids (XSS ).5 0.4 1390.7 68.7 3178.7 57. The feed to the settler was into the 7th layer from the top. . The design parameters related to the activated sludge plant were chosen according to the basic principles of wastewater treatment plant design (Tchobanoglous & Burton.0 16.1 0.8 g m−3 in 20 days (Fig. heterotrophic bacteria (XH ) and inert particulate (XI ).6 282.4 3114.1 to 68.0 30.9 Fig.0 0.3 – 6.0 Aeration tank after 20 days 2.7 g m−3 (Fig. Dissolved oxygen concentration and nitrate nitrogen concentration vary continuously due to the on/off strategy applied to the aeration devices. 4.6 30.0 30. Berber / Computers and Chemical Engineering 30 (2006) 490–499 493 velocity model were used as given by Tak´ acs for low load feeding. The heterotrophic bacteria concentration increases from 30 to 1414.0 100.0 0. Balku.0 0. 3). The aim was to maintain the required microorganism concentration for convenient treatment and proper settling characteristics.1 0.9 Efﬂuent from settler – 30.7 154 7431. The changes in concentrations of dissolved oxygen.5 129.8 124.0 0.
6 1. i.2 4. This alphabet could be binary digits.3 10 3114. will gain a survival capability through the next generations. integers. real numbers.3 8 145. van Straten. 4. called the ﬁtness value.5 0.1 2 4.7 9 144.1 3.1. ‘Parent’ chromosomes are selected by a certain procedure (selection).7 154.5 13. ranking methods. & Guterman.4 0. However. The genetic operators are the basis of the EA search mechanism.6 0.0 7. Berber / Computers and Chemical Engineering 30 (2006) 490–499 Fig.3 0.5 129.3 0.e. Furthermore.0 147. It is shown that the more natural representation to a real number problem.494 S. which are applied on individuals in a population (Ronen. linearity and continuity). the next attempt was devoted to determining the optimal aeration proﬁle using the evolutionary algorithm.9 1.2 8. no waste sludge was removed from the system.3 5 11.4 0.2 345.2 5. symbols.7 7 144. Then the genes of the parents are mixed and reassembled to create their offspring. 4.6 27. 2002). much computational time is lost in conversion of the bit strings to real Table 2 The simulated concentrations of the particulate components in the layers of settler Settler layers/particulates (g m−3 ) XI XS XH XSTO XA XSS 1 3. Evolutionary algorithm After the continuous operation dynamics were identiﬁed. A chromosome represents each individual in the relevant population of solutions.9 6.4 17. The state variables scope is scanned by means of evolutionary resembling operations (selection. A positive value.5 0.4 0. The evolutionary algorithm is one of the stochastic search methods.4 0. ammonium (SNH ). Hessians.2 10.0 7431.5 0. Each upcoming generation is created from the chromosomes of the current pool.7 0.3 g m−3 SSeff : 8.5 1.5 13. The main operators are mutation and crossover. elitist models.6 282.4 0. scaling methods and tournament methods. General Evolutionary algorithms (EA) are based on Darwin’s natural selection principle. 1999). reﬂects the ‘goodness’ of the chromosome and is tightly related to the value of the objective function. is more efﬁcient and produces better solutions space while combinatory problems are better solved using binary representation. Roubus.0 7.2 6.2 345.0 147. In a practical EA problem.1 28. R.2 345.1 2.4 65.9 . boundary and Gaussian (Michalewicz’s real-code genetic operators.2 3 5. but randomly sample the search space. It is expected that the better chromosome will create more offspring and hence. Balku. etc.0 7.8 6 27.1 g m−3 These results show that the above-mentioned system operates in conformity with the European Environment Commission (EEC) discharge criteria (EU.3 7.9 4 7. non-uniform.1 0. high precision numerical problems. matrices. real numbers.5 13. crossover and mutation).9 6.3 0.6 g m−3 TNeff : 7. The binary representation used in classical genetic algorithms (GA) has some drawbacks when applied to multidimensional. stochastic search approaches do not use these features. 4. Twenty days simulation proﬁles for dissolved oxygen (SO ).5 11. a pool of chromosomes is set. Traditional deterministic search methods ﬁnd the next sampling point using the problem features (such as gradients.7 3178. The genetic operators used in crossovering may be arithmetical or heuristic. The bit-strings become very long and the search-space is blown up. Each chromosome is composed of successive genes of a certain alphabet. the surviving chromosome is considered as the best solution. The initial population provided for the EA is usually a random population. During the course of the evaluation. The original genetic algorithm uses binary digits. while the crossover mixes two individuals and creates two new ones. nitrate nitrogen (SNO ) and alkalinity (SHCO ). Some selection functions are: the roulette wheel. The concentrations of the particulate components of ASM3 in the layers of the settler at the end of 20 days start-up operation are given in Table 2. After the operation period of 20 days the efﬂuent water (from the top layer of the settler) has the following characteristics: CODeff : 37. Stabtai.0 147. The mutation changes an individual to create a new one. They create new solutions based upon the current population.3 12. 1998). The selection function deﬁnes the search space of every generation. & van Boxtel. and in mutation they may be uniform.
2. The population size is a parameter to control the optimization progress of GAs. The advantages of the Deb’s method are: 1. But the presence of constraints signiﬁcantly affects the performance of the evolutionary algorithm like any other algorithm. all they need is an evaluation of the objective function. Berber / Computers and Chemical Engineering 30 (2006) 490–499 495 values and vice versa. 2. Optimization problem Energy optimization is achieved by minimizing the aerated fraction in the total operating time. solutions are compared in terms of objective function values alone and in the third case. evolution strategies do not allow equality constraints since the ratio between the sizes of feasible and infeasible search spaces is zero with such constraints). such as roulette selection. in the ﬁrst case neither the objective function value nor the constraint violation information is used. there is no need to evaluate the objective functions of the infeasible solutions and to compare them with feasible ones. the sum of total aerated time and total non-aerated time.g. penalizing infeasible individuals and constraint handling method proposed by Deb are considered: (I) Rejection of infeasible individuals: In this method. one of the main advantages of evolutionary computational techniques is that they do not have much mathematical requirements about the optimization problem. The domain of objective function f is extended as follows: eval(p) = f (p) ± Q(p). Any feasible solution is preferred to any infeasible solution. Deb’s method uses simulated binary crossover (SBX) and a parameter-based mutation. for many search problems where the initial population consists of infeasible individuals only. The objective function is then deﬁned as the ratio of two quantities: total aeration time divided by total operation time. the individuals that do not ﬁt the constraints are rejected from the population. a niching method is used to maintain diversity among the feasible solutions. for example. Otherwise the method has serious limitations. it must not be too low (lower than 10–20) because too much information will be lost in every generation and as a consequence the algorithm may converge too fast and end. 3. R. Using SBX and parameter-based mutation ensures the new generations remaining in the boundaries. is penalizing infeasible solutions. Michalewicz. There are no ﬁxed rules for the population size when using a GA. In his paper. penalty parameters are not needed because in any of the above three scenarios. is to be used together with the tournament selection proposed by Deb. GA’s population-based approach and ability to make pair-wise comparison in tournament selection operator are exploited to devise a penalty function approach that does not require any penalty parameter. Instead of calculating the objective functions of the infeasible solutions the author chooses to place the maximum objective function value among the feasible solutions in the current population. However. However. where two solutions are compared at a time always enforcing the following criteria: 1. Le Riche. This offers a few simpliﬁcation of the algorithm. For example. The other main efﬁcient features proposed by Deb are the usage of the simulated binary crossover and the parameter-based mutation. (III) Deb’s method: The third method proposes to use a tournament selection operator. solutions are compared in terms of constraint violation information alone. & Schoenauer (1996) focused on and classiﬁed the constraints handling methods in evolutionary computation techniques. etc. Presence of the constraints From the optimization viewpoint. Among two infeasible solutions the one having smaller constraint violation is preferred. Dasgupta. solutions are never compared in terms of both objective function and constraint violation information. One general way of dealing with constraints. In the proposed method. Once sufﬁcient feasible solutions are found. Generally. In the present paper rejection of infeasible individuals. For the real-coded GA. Such a method may work reasonably well when the feasible search space is convex and constitutes a reasonable part of the whole search space (e. min J = M k k=1 b M k k k=1 (a + b ) (8) . and the evaluation of the maximum objective value for the feasible solutions is preferred only when the evaluation of the objective functions spends an important CPU time in the solutions and only if any other selection method depending upon the ﬁtness values. No need for the calculation of the penalty parameters in the selection procedure. Since this method does not require any penalty function approach. there are no guidelines on designing penalty functions. simply the feasible solution is preferred. 4. the one having better objective function value is preferred. Of the three tournament cases mentioned above. Balku. (II) Penalizing infeasible individuals: This method is the most common approach in the genetic algorithms community. In the second case.S. 2. weighing the constraints continues to be a question. no solution is obtained. The major difﬁculty in this method is the estimation of the penalty parameters. whatever the optimization method is. it is easy to apply any kind of constrained problems. An efﬁcient constraint handling method is proposed by Deb (2000). 5. where Q(p) represents either a penalty for infeasible individual p or a cost for repairing such an individual. Among two feasible solutions.
. it is known that the DO concentrations are generally maintained. 3. the deviation from discharge standards and average DO at the end of the aeration periods are calculated. respectively. TNeff − TNmax ) + max(0. Therefore. 2000). 2. In order to overcome this difﬁculty. Such a constraint is not involved in the current literature (Chachuat et al. The application of the GA operators is the same in the ﬁrst and second methods but differs in the third method. . 2. and 60 min for non-aeration and 120 min for aeration periods as maximum. 6. 4. since the overall economical objective is to minimize the aeration period. Penalizing infeasible individuals In this solution algorithm. and ﬁnd percentages in order to ﬁnd the ﬁtness values. . such low levels of DO concentrations are not suitable for the activated sludge systems including nitriﬁcation because the activities of nitrifying bacteria decrease at low levels of dissolved oxygen. tournament selection (Section 4. 2001). 6. in contradiction to the practice. a penalty function is added to the objective function in order to calculate the ﬁtness value before roulette The minimum and maximum lengths of non-aeration and aeration sequences are. around (though not strictly) 2 g/m3 in practice. The deviation from the efﬂuent discharge criteria was formulated as: Error = max(0. For above-mentioned reasons such a limitation is added to the optimization algorithm.1. an additional constraint is added to the algorithm and the DO concentrations at the end of the aeration periods are expected to be controlled by holding the average DO concentration at the end of the aeration periods to be above 2 g m−3 .2. CODeff − CODmax ) + max(0. All of the methods use real numbers and the same solution approach. Arceivala deﬁnes a DO concentration range between 1. Mutation is applied in order to prevent the chromosomes repeating themselves (whole uniform mutation). Solution algorithm The optimization problem is subject to a set of differential equations. Running the global model for each individual. Each interval (k) was composed of a pair with one non-aeration period and one aeration period. Balku. Our simulation studies show that if DO concentration is not constrained. Mutation: A randomly selected gene of a randomly selected chromosome is replaced by a value randomly given but in compliance with the upper and lower boundaries. SO varies between 0 and 2. the objective function. 6. ASM3 works at low dissolved oxygen concentrations.0 mg/l for proper nitriﬁcation . The constraints on the system are the efﬂuent discharge criteria and the dissolved oxygen concentration. The objective function is subject to the mass balances around the aeration tank and the settling tank such as represented in Eqs.2). 2004). Since the aerobic and anoxic modes are both considered in the same tank. a waste sludge stream of 2% of inlet ﬂow rate was introduced to the settler model. CODeff and SSeff are calculated by their deﬁnitions in ASM3. only the infeasible solutions are rejected during the computation. In the third method. Solution algorithms for the deﬁned optimization problem by three different methods of handling constraints are given in this section. simulated binary crossover and parameter-based mutation are applied (Deb. Rejection of infeasible individuals The ﬁrst solution algorithm is based upon the general evolutionary algorithm with the known GA operators. 5..5 and 2. which represent the process behavior and has some hard constraints on the system.6 g m−3 . M. 1986). By roulette wheeling rearrange the individuals. the genes of the chromosomes are replaced by simple arithmetical crossover. whose durations in time scale were ak and bk . R. SSeff − SSmax ) (9) where TNeff . Berber / Computers and Chemical Engineering 30 (2006) 490–499 The time scale during which the optimum sequence of nonaeration/aeration periods to be found is discretized with M intervals such that k = 1.496 S. (6) and (7). The ﬂow chart of the method for the rejection of the infeasible individuals is shown in Fig. instead of rejection of the infeasible individuals. The required efﬂuent criteria are taken as: TNmax : 10 g m−3 CODmax : 125 g m−3 SSmax : 30 g m−3 According to our numerical experiments the most important part of the optimization is to control the dissolved oxygen concentration in the aeration tank. In the simulation of continuous operation.. it seems that a successful treatment may be achieved also at low SO values during aerating periods. A certain percentage of the individuals is chosen for the future operations where the chance to be chosen is proportional to the goodness of the individual which depends upon the objective function. 15 min for both sequences as minimum. Therefore. the DO levels may decrease under the limits deﬁned in literature in order to meet this criterion (Balku. Elitism: The individual having the best ﬁtness function value in the selection is added to the new generation without crossover and mutation. Take their inverse. Obviously. Selection: Arrange objective functions in ascending order. The boundary conditions for the aeration and nonaeration sequences are: on on tmin ≤ bk ≤ tmax off off tmin ≤ ak ≤ tmax (Arceivala. Crossover: The chromosomes are paired randomly and then starting from a random position. The algorithm in the application of the GA operators for the ﬁrst and second methods is as follows: 1.
07 and 54. Give real.65 Fig. 7. 5.S. A chromosome represents an aeration proﬁle consisting of non-aerated and aerated periods. no of iteration Objective function Constraint violation Total computational time (h) Rejection of feasibles 1000 30 50 55. The values are Fig. Alphas (α) are the penalty parameters. Results and discussion The results obtained by the three different optimization methods dealing with constraint handling are shown in Table 3. Q(p) = β[α1 abs(Tcalc − Treq ) + α2 abs(error) + α3 abs(DOcalc. selection. 7. random values for each gene in all chromosomes in the population. Deb’s method In the third solution algorithm. 6. 2 M) in a chromosome. 6. the values of the objective functions increase while the solution individuals are moving far from the constraints. zero is given to penalty parameters.00 Penalizing feasibles 21 30 100 58. R. 10. The best result is obtained by Deb’s method in terms of objective function in a predeﬁned maximum iteration number as it can be seen in Table 3. replace the disconformities with the upper and lower boundaries.3. if not. Mutation is applied according to the parameter-based mutation. (6)) forward in time starting from X (t0 ) = X0 . Balku. penalizing and Deb’s method. Choose the number of population and the number of genes (an even number. The penalty function consists of some linear equations deﬁning violations from the constraints. 3.01 by rejection. the only difference is the placement of real objective function values of the infeasibles in the ﬁtness function). Control whether upper and lower limits are satisﬁed.2. instead of roulette selection. Assign the initial values (X0 ) calculated by the start-up simulation for the state variables.07 – 65.04 – 68. and integrate the aerated model (Eq. Evaluate the objective functions (this evaluation takes nearly no CPU time) and the constraints (can only be evaluated by running the model and most of the CPU time is spent for this evaluation because of the differential equations). 58. 6. 9. a function of the violations from the constraints and of the parameters deﬁned according to the degree of violation. The changes in objective function by iteration numbers for the rejection of the infeasible are shown in Fig. Berber / Computers and Chemical Engineering 30 (2006) 490–499 Table 3 The results obtained by the application of the proposed methods Method applied Initial number of population Number of variables Max. tournament selection is applied based upon the method proposed by Deb. ﬁnd the values of Xk at the end of the time period of ak .36 497 Deb’s method 20 30 50 54. (7)) forward in time starting from Xk for the time period of bk and so on until it is applied to the each gene in one chromosome and for all the chromosomes in the population. The objective functions evaluated are 55. If the constraints are satisﬁed to an acceptable extent. Go to the 5th step in order to apply the GA operators to the new generation. Tournament selection is performed by Deb’s method (deﬁned in Section 4. Integrate the non-aerated model (Eq. .ave − DOreq )] where β is the weighing factor between objective function and penalty function which was chosen as 9. Therefore. Otherwise they are chosen as 10. 8. 2. Flow-chart for rejection of infeasible individuals handling algorithm.01 – 34. respectively. Changes in objective function (rejection). The suggested algorithm: 1. SBX method is used in crossover. 5.04. 6. The application of the penalty function is as follows: Q(p): the penalty function. 11. Continue the iteration until a predeﬁned maximum iteration number is met. 4.
if a high population number is selected. In aerated periods SNO concentration increases. total suspended solid (SSeff ) and total nitrogen (TNeff ) as a function of time by the variable optimized proﬁle. On the other hand. Changes in chemical oxygen demand (CODeff ). 6. since the GA operators acting on remaining limited ones cannot considerably change the results in proceeding iterations. This algorithm eliminates the chromosomes (individuals). 7. whereas the white bars indicate those of the aerated periods. until another elite chromosome directing towards a better objective function is found. For this reason. which is the desired value. the most important difﬁculty in penalty function approach and further using SBX and parameter-based mutation results in better solutions in the next generations. The variable optimized proﬁle obtained by Deb’s method is shown in Fig. The third method that is proposed by Deb has some advantages in comparison to the previous methods. on the other hand. Changes in objective function (penalizing). The suggested second algorithm contains an augmented penalty term added to the objective function. The black bars indicate durations of the non-aerated periods. and 18. 10.98% energy is saved by the optimal proﬁle in comparison to a constant arbitrary proﬁle (0. The maximum number of iterations is very important in this case. 8. the parent population should be chosen as much as large. because if it is not big enough. Optimal aeration proﬁle using Deb’s method.9 h non-aerated and 1. 8. nitrate plus nitrite nitrogen (SNO ) concentrations as a function of time by the variable optimized proﬁle. ammonium plus ammonia nitrogen (SNH ). The method does not require the evaluation of any penalty parameters which is Fig. Changes in dissolved oxygen (SO ). Keeping the elite chromosome can maintain the objective function constant for a number of proceeding iterations. a feasible population may not be achieved. How big a maximum number of iterations is selected is not so important in this algorithm in comparison to the population size. as reﬂected in Fig. 7. The changes in nitrogen compounds and the dissolved oxygen concentrations in time are shown in Fig. the solution population may consist of only infeasible individuals.498 S. The same problem related to the high CPU times is also valid in this method anyhow due to the features of the problem deﬁned in this paper. Fig. which do not satisfy the constraints. when a large maximum number of iterations is chosen. 9. the solution requires high CPU times. As it is seen from Table 3 using the penalizing feasibles method the penalty value has decreased to zero. If it is a small number. . The changes in the ﬁtness function with respect to the iteration numbers are shown in Fig. 9. the drawback is the increased computational time. up to the 45th iteration since the number of individuals decreases to one at that iteration and the program is automatically stopped. R. In the solution generation 70% of the population is in the feasible region.8 h aerated). and contrarily SNH concentration decreases Fig. Berber / Computers and Chemical Engineering 30 (2006) 490–499 Fig. and thus the values of the ﬁtness function may become occasionally large for the infeasible solutions. The values obtained from the suggested optimization algorithm have satisﬁed the constraints. Balku.
0005 h in ode15s integration routine in all three methods. 25(10). Ankara University. 535–544.44.. W. McGraw-Hill. thesis (in Turkish). Dynamic optimization of small size wastewater treatment plants including nitriﬁcation and denitriﬁcation processes. Zhao. int/. D. G.4 GHz Pentium 4 processor having 512 MB memory were used for all computations. Henze. 1263– 1271. High computational times cause hesitation to search global minima and the attempts to ﬁnd the global optima are left for the future studies. Berber / Computers and Chemical Engineering 30 (2006) 490–499 499 depending upon increasing SO .. (2002).. Y. 10 showing that at any time the discharge criteria are met. J. I..S. C. 186. M. 12. The energy savings achieved by the variable optimized proﬁle obtained by Deb’s method in comparison to the constant optimized proﬁle will be 5. 1. Water Science Technology. the proposed evolutionary solution methods to the constrained dynamic optimization problem of an activated sludge system show that denitriﬁcation can also be achieved in the aeration tank by an on/off strategy applied to aeration devices determined by optimal aeration proﬁle.81%. (2004). G.. It is seen that all the methods give satisfactory results in terms of the required treatment despite none of them being global optima and energy savings to the extent of 17. Computers and Chemical Engineering. M. p. An evolutionary strategy for fed-batch bioreactor optimization. Tchobanoglous. 67. Hao. Deb. R. MATLAB® 6. London: IAWQ.34 and the optimized constant proﬁle consists of 1.. Additionally.. References Arceivala. (1999). (1991). Alternating aerobic–anoxic process for nitrogen removal: Dynamic modeling. 253–263. M. . D.90 and 18.00 h for non-aerated periods and 1.. 8. Wastewater engineering: Treatment. Henze.. are possible in comparison to a constant arbitrary proﬁle. M.5 software package and a 2.. (1999). Consequently. A. New Delhi: Tata McGraw-Hill Publishing.98% for rejection.. M. S. Roche. one run is executed by Deb’s method in order to ﬁnd an optimized constant proﬁle. Isaacs. Gujer. O. 187). 39(1).. http://europa. 25. S. & Huang. Marais. 311–338. Water Environment Research. J.81% in comparison to the constant optimized proﬁle.eu. (1986). Balku. & Matsuo. Tak´ acs. Journal of Biotechnology. Optimization of feeding proﬁle for a fed-batch bioreactor by an evolutionary algorithm. (1995). J. 68(1).. M. 173–187.. Michalewicz. (2001). (1996). (1987). Le Riche. R.D. Mino. G. (1991). The average DO concentration at the end of the aeration periods is 2. the dynamic start-up simulation of an activated sludge system including nitriﬁcation and denitriﬁcation has been pursued in order to evaluate the normal operating conditions. Alternating aerobic–anoxic process for nitrogen removal: Process evaluation. Conclusion In the present study. & Guterman. Water Research. Stabtai. L. T. J. Water Research. (2000). & Loosdrecht. A. disposal and reuse (3rd ed. Computer Methods in Applied Mechanics and Engineering. H.. J. W. Dasgupta. (1996).01 g m−3 and the deviation from the discharge criteria is zero. H. 183–193.. 30(4). It is to be noted that the best objective function can be achieved is a variable period optimum proﬁle by Deb’s method of constraint handling. O. & Latiﬁ. 68(1). Burton. Huang. SSeff and TNeff changes given in Fig. Chachuat. A dynamic model of the clariﬁcation—Thickening process. penalizing and Deb’s methods. J.34 h for aerated periods. Activated sludge model no. & van Boxtel. 94–104. Gujer. N. Balku.. & Schoenauer. B. Energy savings could be possible to the extent of 18. 83–93. Energy optimization in activated sludge systems with nitrogen removal. Ronen. 585–593. Soeberg. K. Computers & Industrial Engineering. Department of Chemical Engineering. Wastewater treatment for pollution control (p. & K¨ ummel. 851–870. Evolutionary algorithms for constrained engineering problems. M... Grady. G. G. A. Activated sludge model no. but all of them have suffered high computational times due to the integration of the large set of differential equations at every stage of the solution algorithm. 1. Scientiﬁc and Technical Reports No. The operation by optimal aeration proﬁle results the CODeff . G.. Ph. Roubus.. & Hao.. H. The other features that may have contributed to the large computation time requirement are constraints. & Nolasco. 97. T. such as the deviations from discharge standards and DO concentration that must be controlled. F. P. van Straten. Patry.. 3. Solution of the optimization problem has taken quite substantial computation time probably because of the complicated dynamic model and the maximum step size of 0.. B. respectively. concepts and performance. Water Environment Research. The results show that the value of objective function is 57.98% in comparison to a constant arbitrary aeration proﬁle and 5. An analysis of nitrogen removal and control strategies in an alternating activated sludge process. (1996). Then the determination of the optimal aeration proﬁle for the system has been accomplished by the evolutionary algorithm using three different constraint handling methods. 29(2).. 1334). J. EU-Directive 98/15/EEC amending directive 91/271/EEC. Journal of Biotechnology. Z. An efﬁcient constraint handling method for generic algorithms. H.
This action might not be possible to undo. Are you sure you want to continue?